Deep Learning Essentials

Visual Geometry Group

This architecture from Simonyan and their co-authors [17] was the runner-up in the ImageNet challenge in 2014. It is designed on the core idea that deeper networks are better networks. Though they provide a higher level of accuracy, they have an inherently larger number of parameters (~140M) and use a lot more memory than AlexNet. Visual Geometry Group (VGG) has smaller filters than AlexNet, where each filter is of size 3 x 3 but with a lower stride of one, which effectively captures the same receptive field as a 7 x 7 filter with four strides. It has typically 16-19 layers depending on the particular VGG configuration. The VGG CNN architecture figure illustrates this architecture:

VGG CNN architecture (Source: https://raw.githubusercontent.com/PaddlePaddle/book/develop/03.image_classification/image/vgg16.png) ...